Multi-Phase CT-Based Radiomics Nomogram for Discrimination Between Pancreatic Serous Cystic Neoplasm From Mucinous Cystic Neoplasm

Abstract

PurposeThis study aimed to develop and verify a multi-phase (MP) computed tomography (CT)-based radiomics nomogram to differentiate pancreatic serous cystic neoplasms (SCNs) from mucinous cystic neoplasms (MCNs), and to compare the diagnostic efficacy of radiomics models for different phases of CT scans.Materials and MethodsA total of 170 patients who underwent surgical resection between January 2011 and December 2018, with pathologically confirmed pancreatic cystic neoplasms (SCN=115, MCN=55) were included in this single-center retrospective study. Radiomics features were extracted from plain scan (PS), arterial phase (AP), and venous phase (VP) CT scans. Algorithms were performed to identify the optimal features to build a radiomics signature (Radscore) for each phase. All features from these three phases were analyzed to develop the MP-Radscore. A combined model comprised the MP-Radscore and imaging features from which a nomogram was developed. The accuracy of the nomogram was evaluated using receiver operating characteristic (ROC) curves, calibration tests, and decision curve analysis.ResultsFor each scan phase, 1218 features were extracted, and the optimal ones were selected to construct the PS-Radscore (11 features), AP-Radscore (11 features), and VP-Radscore (12 features). The MP-Radscore (14 features) achieved better performance based on ROC curve analysis than any single phase did [area under the curve (AUC), training cohort: MP-Radscore 0.89, PS-Radscore 0.78, AP-Radscore 0.83, VP-Radscore 0.85; validation cohort: MP-Radscore 0.88, PS-Radscore 0.77, AP-Radscore 0.83, VP-Radscore 0.84]. The combination nomogram performance was excellent, surpassing those of all other nomograms in both the training cohort (AUC, 0.91) and validation cohort (AUC, 0.90). The nomogram also performed well in the calibration and decision curve analyses.ConclusionsRadiomics for arterial and venous single-phase models outperformed the plain scan model. The combination nomogram that incorporated the MP-Radscore, tumor location, and cystic number had the best discriminatory performance and showed excellent accuracy for differentiating SCN from MCN.

Radiomics advances in the evaluation of pancreatic cystic neoplasms

“…A nomogram containing clinical features (number of cysts) and the three-phase radiomics features of plain scan, late arterial phase, and venous phase were created, and the AUC of the model in the validation cohort was 0.817, which was higher than that of the models built with single-phase or dual-phase CT features. Similar findings were also reported by Gao et al [ 8 ], who also found that radiomics models based on the venous phase or arterial phase were more effective than those based on the plain phase. Shen et al [ 42 ] selected five radiomics features and four clinical indicators (serum CA19-9, CEA, sex, and age) with the Boruta method and built multiclassification diagnostic models with SVM, RF, and ANN.…”

“…For patients with low-risk branch duct IPMN (BD-IPMN) and MCN patients with a tumour size less than 4 cm who are asymptomatic or low risk, guidelines also recommend observation follow-up [ 5 ]. The accuracy of conventional imaging diagnosis of PCN is less than 50 % [ [6] , [7] , [8] ]. Improvement in the diagnosis accuracy and risk assessment of PCNs is still a major clinical challenge, and new methods are urgently needed for the evaluation of PCNs.…”

Pancreatic Incidentaloma

“…CT-based radiomics is an emerging and rapidly developing method for advanced image analysis and comprises feature extraction of various phases (plain scan, arterial phase, vein phase) of contrast-enhanced CT. Next, this action is used to construct a combination normogram incorporating a multi-phase radiomics model, involving the extraction of all three phases of CT to discriminate non-invasive subtypes of PCN. Gao and co-workers revealed that radiomics for arterial and venous single-phase models outperformed the plain scan model, and a combination nomogram that incorporated the MP-Radscore, tumor location, and cystic number had the best discriminatory performance and demonstrated excellent accuracy for differentiating SCN from MCN [ 76 ]. In addition, multi-phase post-CT radiomic evaluation could improve predictive capability in diagnosis of the malignancy in IPMNs and has benefits over single venous phase CT analysis [ 77 ].…”